[OpenID] Why use SREG instead of AX?

[John Bradley]

Breno,

It's on my list for this weekend.

I think Andrew's idea is a interesting solution.

I don't think that overloading the <Type> attribute is a good long  
term solution.

I hope we could do something better for XRD 1.0 discovery to describe  
the features of a OP.

In the short term overloading <Type> is something we can do.

The other pressing issues for AX are:
1 Registering a <Type> for discovering the RP privacy policy
2 Registering a <Type> for discovering the RP Terms of Service.
3 Having a way for the RP to know if the OP supports if_available.

The last one is important to deal with the fact that some OP ignore  
if_available in requests and only give the user the choice to return  
the attributes that are requested as "REQUIRED".

All the OPs could agree on a single way to interpret AX requests but  
that is a touch optimistic:)

Having a way for a RP to differentiate between OPs operating on  
different theories would be good for RPs.

John B.
On 10-Jul-09, at 1:35 PM, general-request at openid.net wrote:

> Date: Fri, 10 Jul 2009 10:00:05 -0700
> From: Breno de Medeiros <breno at google.com>
> Subject: Re: [OpenID] Why use SREG instead of AX?
> To: Andrew Arnott <andrewarnott at gmail.com>
> Cc: general at openid.net
> Message-ID:
> 	<29fb00360907101000n77f57032wdd0635a60b660ff at mail.gmail.com>
> Content-Type: multipart/alternative;
> 	boundary=0016369205b372e192046e5ce7b4
>
> --0016369205b372e192046e5ce7b4
> Content-Type: text/plain; charset=ISO-8859-1
> Content-Transfer-Encoding: 7bit
>
> For one, it would be interesting if someone with expertise on how to  
> create
> lightweight processes for registration would clarify the language in  
> that
> wiki document. Hopefully, it would be possible to define a process  
> that
> exempts the OIDF from the burden of having to impose the current IPR  
> policy
> on each contributor. (I had hoped JBradley would take that on, but  
> he has
> plenty on his hands at this point).
> Once that wiki page is in good shape we could spam specs@ to bless  
> it and
> then we can get started right away. There is even an initial  
> understanding
> to use the domain name schemas.openid.net  and to host each  
> registered URL
> at that domain with a description of the wire format for the  
> attribute. So
> once this is up, I think progress could be made quickly.